Linking to GEOS 3.13.0, GDAL 3.8.5, PROJ 9.5.1; sf_use_s2() is TRUE
Attaching package: 'dplyr'The following objects are masked from 'package:stats':
filter, lagThe following objects are masked from 'package:base':
intersect, setdiff, setequal, unionThe aim of this note is to examine the different statistical and cartographical sources of interest for the objectives of the REDEC project. We have chosen the département of Val-de-Marne as example because of this exceptional complexity in terms of contradictions between the different geographical divisions that should be used in the analysis
The most fundamental units for the project are cantons and cicronscription. In the case of urbanised area like Val-de-Marne,it is not possible to build cirnconscription without breaking some cantons. But what is the level of discrepancy between the two divisions ?
The elaboration of 11 circonscription with 25 cantons as building block should normally be feasible by simple agregation with a minimum of cases of fragmentation of cantons. But it is not at all what is observed in the département of Val-de-Marne.
In the french electoral system, the communes are a crucial scale of governance below the level of departement (electtion by cantons), the region (election by département) and the national assembly (election by circonscription). It is therefore not possible to understand the logic observe at the levels of cantons and circonscription without looking at the interactions with the level of commune. The differents scales of the electoral system are strongly interlinked.
In many cases, the delineation of cantons has divided the most populated communes in subparts. Was it always justified ? And what was the political resason of such subdivisions ?
Admitting that it was necessary to divide some communes at the level of cantons, wasn’t it possible to proceed to their reunification at the level of circonscription ? It should normally be possible but it is not always the case as we can see on the map below.
The evaluation of political strategies that has been used by the author of territorial delineation of cantons and circonscription can not be clarified by data collected at communal level in the urbanized area. To analyze social homogeneity of population in terms of social and economic characteristics or political orientations, we need informations at a lower scale. But the great problem for this task is the fact that socio-economic data and political data are not available on the same territorial mesh.
Since many years, the french census offers extraordinary precise data at the level of IRIS which are spatial blocks of population with a minimum of approximatively 2000 inhabitants. Such data covers all aspect of social and ecomic life and are associated to many tables available on the website of INSEE. And they can even be related to individual databases where the location of household and people can be precisely associated to the IRIS. It is therefore very simple to analyze the internal social characteristics of circonscription and produce related maps.
The votation office are places where citizens are going to make their votations. It is generally a primary school or another municipal infrastructure that play the role of office. The size is a bit smaller than IRIS (approx. 1000 citizens per office). The main problem is that the precise geography of votation office is difficult to represent on a map because it is a list of streets and adresses and not a polygon with borders. Many researchers has tried to indentify theses limits and to cross the area of votation office with the IRIS in order to compare socio-economic data and electoral choices. But it remains difficult and subject to approximations. It is nevertheless of very high interest as we can demonstate by two simple examples.
The evaluation of the geometric properties of territorial divisions in terms of compacity or accessibility could be analyzed with the infra-communal data at IRIS level. But the best solution is to used the grid data provided by INSEE at the level of 1km or even 200m. These grid data extract from the census provide information not only on population but also on age, social categories and even income of inhabitants. It can be aggregated at upper level and could to some extent help to solve the problems of correspondance between IRIS and votation offices. These grid data can bu used for geometric measure based on euclideand distance. But they can also be combined with accessibility data by road or public transportation for more precise analysis if needed.
Lets just take an example with another circonscription of the Val-de-Marne localized in the outer periphery with a large superficy
A static map is a projection in euclidean space making possible to compute easily geometric parameter (gravity center, ellipsoid, potential, etc.).
Warning: attribute variables are assumed to be spatially constant throughout
all geometriesWarning: st_centroid assumes attributes are constant over geometries
A dynamic map is a non projected map that can be combien with tiles making more easy to analyze the distributions of population in relation with topographic information, network, etc…